Using station data for bias correction of ground-level ozone concentrations

Ozone is a strong oxidizing agent which makes it one of the most important pollutants in the troposphere. Its high concentrations can be harmful for fauna and flora, therefore it is necessary to monitor its climatological behavior. Air quality stations measure ozone concentrations at singular points and so to obtain the information over an area, chemistry-transport models (CTMs) can be used, which generate gridded fields of pollutant concentrations. Previous studies have shown however, that CTMs are sensitive to boundary conditions (BCs). If those BCs were generated using a global chemistry model, they may introduce additional biases. This represents a problem, since future concentration projections often use this type of BCs. Different ways to correct biases via postprocessing have been introduced in literature using station data, but such methodology may perform poorly when applied to simulations of high horizontal resolution. For this reason, we propose a new method that interpolates station quantile biases when compared to the model simulation. This kind of interpolation takes into account both spatial distribution of the concentrations within the domain as well as the statistical distributions within each model grid-cell. We used station data from Eionet to postprocess WRF-Chem and CAMx simulations utilizing different methods of bias correction. Our experimental setup consisted of the model domain of central Europe with horizontal resolution of 9 km. The simulations were conducted in the 10-year period of 2007-2016 with output timestep of 1 hour. The chemistry BCs of these simulations were taken from the global model SOCOLv4. Our current results show that our approach outperforms the ones found in literature, highlighting the possibility of using station data to correct climate projections of high resolution.